Synthesis and Characterization of Lignin Hydrogels for Potential Applications as Drug Eluting Antimicrobial Coatings for Medical Materials.

Lignin is the second most abundant biopolymer on the planet. It is a biocompatible, cheap, environmentally friendly and readily accessible material. It has been reported that these biomacromolecules have antimicrobial activities. Consequently, lignin (LIG) has the potential to be used for biomedical applications. In the present work, a simple method to prepare lignin-based hydrogels is described. The hydrogels were prepared by combining LIG with poly(ethylene glycol) and poly(methyl vinyl ether-co-maleic acid) through an esterification reaction. The synthesis took place in the solid state and can be accelerated significantly (24 vs 1 h) by the use of microwave (MW) radiation. The prepared hydrogels were characterized by evaluation of their swelling capacities and with the use of infrared spectroscopy/solid-state nuclear magnetic resonance. The prepared hydrogels showed LIG contents ranging between 40% and 24% and water uptake capabilities up to 500%. Furthermore, the hydrophobic nature of LIG facilitated loading of a model hydrophobic drug (curcumin). The hydrogels were capable of sustaining the delivery of this compound for up to 4 days. Finally, the materials demonstrated logarithmic reductions in adherence of Staphylococcus aureus and Proteus mirabilis of up to 5.0 relative to the commonly employed medical material poly(vinyl chloride) (PVC).


Lingnin characterization
The chemical composition of the used lignin is presented in Table S1. The dealkaline lignin was used as received, so it was initially unwashed. This sample showed a value of total lignin content around 80%, which is commonly considered the sum of Klason lignin (KL) and acid soluble lignin (ASL). The value of ASL was remarkably high, which is probably due to the severe cooking conditions of the employed pulping process, which would increase the proportion of low-molecular-weight lignin 1,2 .
The number of inorganic particles (14.72%) is consistent with the results obtained by other researchers using the same type of lignin 3,4 . Part of these inorganics is the sulphur content found in the elemental analysis (2.34%) (Table S2). Also, the analytical techniques of ICP-OES and SEM-EDAX revealed the presence of Na and S 3,5 . All this, therefore, indicates that the used lignin might have been isolated performing a kaft process, since in this pulping method Na 2 S and NaOH are used in different proportions as reagents 5 . The carbon and hydrogen contents (Table S2) are in the range of other technical lignins. In lignin samples, it is assumed that the Nitrogen content comes from cell proteins. The lignin used in this work, showed a Nitrogen (Table S2) content similar to wood lignin samples, which is much lower than those obtained from wheat straw lignin 6 . Finally, the absence of glucose in the carbohydrates analysed confirms that the cellulose is devoid of cellulose contamination (Table S1).  Figure S1. FTIR spectra of the LIG sample The chemical structure of the employed lignin was studied using FTIR and Solid State 13 C-NMR. The FTIR of the lignin was recorded ( Figure S1) and compared with the assignments found in other scientific papers [6][7][8] . As observed form the Figure S1, the dealkaline lignin shows a wide band centred at 3372 cm -1 corresponding to the aromatic and aliphatic OH groups. Moreover, bands located 2937 and 2850 cm -1 are assigned to the symmetrical and asymmetrical C-H stretching of the methyl and methylene groups, respectively. A band representing the asymmetric deformation of C-H stretching also appears at 1456 cm −1 . Additionally, the absorption bands located at 1592, 1511, and 1423 cm -1 were assigned to aromatic ring vibrations of the phenylpropane units. The observance of a band at 1649 cm -1 is due to stretching vibrations of conjugated carbonyl groups. The band at 1321 cm -1 can be assigned to C-O stretching in the syringyl units, which is typical for non-wood lignins 2 . Although the supplier does not give so much information about the pulping process or the raw material used, this band was practically inexistent in the used lignin sample which could indicate that a wood species was employed. Furthermore, bands located at 1267 and 1213 cm -1 are associated to guaiacyl units, indicating G ring and C=O stretch. Also, the intense band located at 1033 cm -1 could be assigned to stretching vibrations of C-H bonds in the guaiacyl structure. This band is less intense or even inexistent when the sample has a lower amount of G units 6 . The most intense band observed at 1123 cm -1 is assigned to C-O stretching in ether and alcohol groups. Finally, the band at 617 cm -1 was attributed to C-S stretching 7 . It also indicates that the lignin was isolated using a pulping process containing sulphur delignifying agents and/or was precipitated employing sulphuric acid. shown in the Figure S2. The peak around 56 ppm has been attributed to the methoxyl groups 9 .
According to the supplier this lignin has 11.1% of these groups (calcd. on anh. substance). The peaks around 74 and 104 ppm could be attributed to the resonance of carbons in carbohydrates moieties 2 and these two peaks are practically inexistent in the used lignin. This support the results found in the  Figure S3A shows the spectra of the pure components and of LIG10K before (LIG10K NC) and after the crosslinking process. The main difference is the formation of anhydride groups between the contiguous COOH groups in GAN 12,13 . However, these new anhydride groups are labile and disappear when the hydrogels are hydrated or over prolonged periods of time due to ambient humidity. A S5 further point of interest concerns the shift in the carbonyl peak (1760-1680 cm -1 ) observed for the non-crosslinked materials when compared with pure GAN ( Figure S3B). In pure GAN, there are noncovalent interactions, mainly hydrogen bonds, between the acid groups 14,15 . When GAN, LIG and PEG were combined, these interactions disappeared because GAN chains were mixed with the other molecules producing a shift in the peak to higher wavenumbers. The films containing GLY instead of PEG displayed a lower peak displacement. This suggests that the presence of GLY did not preclude the establishment of non-covalent interactions between GAN chains. This can be easily explained by the smaller size of the GLY molecule than the PEG molecules used in the present work.  Figure S4A shows the solid state 13 C-NMR spectra for the major pure components of the hydrogels. It can be seen that some of the main peaks are not overlapping and, consequently, they have been used to estimate the sample composition. The peaks between 158 and 100 ppm were used for estimation of the concentration of LIG in the hydrogels. These signals are due to aromatic carbons in LIG and they are a unique identifier for LIG ( Figure S4A) 16 . The peak between 185 and 165 ppm was used to estimate the concentration of GAN in the hydrogels. This peak can be assigned to the acid group carbons of GAN (see Figure S4A) 17 . Finally, the peak between 77 and 66 ppm was used to estimate the concentration of PEG in the hydrogels.

Solid state 13 C-NMR analysis of the LIG-based hydrogels
S6 Figure S4. Solid state 13 C NMR spectra of LIG, GAN and PEG (A) and the hydrogels (B). Figure S4B shows the solid state 13 C-NMR spectra of the LIG14K initial mixture (LIG14K NC) and of the other hydrogels after crosslinking and washing. In order to estimate the component percentages, the previously described peaks were used. The peak area ratio of LIG/GAN (area of the 158-100 ppm peak/area of the 185-165 ppm peak) and LIG/PEG (area of the 158-100 ppm peak/area of the 77-66 ppm peak) was calculated for the LIG14K NC sample and was used as a standard since the LIG:GAN:PEG ratio in LIG14K NC was known (2:1:1). The LIG/GAN and LIG/PEG peak area ratios were calculated for all synthesised hydrogels and compared with the standard. The obtained results are shown in Table 1 in the manuscript. The composition of the LIGGLY hydrogel could not be estimated due to the low intensity and overlap of the peak assigned to GLY at 77-66 ppm with peaks from GAN and LIG. Figure S5. SEM images of the dry and freeze dried (FD) LIG-based hydrogels. Figure S6. CUR cumulative release from all the LIG-containing hydrogels hydrogels (n=3). Figure S6 shows the cumulative CUR release from the LIG-based hydrogels. Additionally, Table S3 shows the results obtained after fitting CUR relase data to different kinetic models. After fitting the release data ( Figure S6) to the Korsmeyer-Peppas model, LIG14K showed an n value close to 0.45, indicating that Fickian diffusion was the main mechanism governing the release process 18,19 . This is consistent with the high correlation obtained after using the Higuchi model (Fickian diffusion model)

Curcumin release and mathematical modelling
for the release data from LIG14K hydrogels. In contrast, the other hydrogels showed lower n values suggesting a burst release. The CUR was rapidly released and did not follow a defined release mechanism. Consequently, these hydrogels displayed poor correlation with the Higuchi model as CUR release from these hydrogels did not follow a Fickian diffusion model. These models were applied to the sections of the release curves up to 60% of the CUR release. However, LIGGLY and LIG10K showed linear release after the initial burst release. Consequently, a Zero-Order model was applied to the linear segments of the curves for these two hydrogels. The results obtained after using a Zero-Order model for the linear range can be seen in Table 4. Both hydrogels showed similar k ZO indicating similar release kinetics. The only factor that can explain this behaviour is the similarities in the swelling profiles between these two hydrogels. However, LIG10K showed a superior release capability as these hydrogels were able to take up higher amounts of CUR.   In Figure S8, the significantly higher resistance of the LIG10K samples pre-soaked for periods of 0 h, 24 h and seven days to adherence of S. aureus and P. mirabilis after 4 h and 24 h incubation relative to GANPEG controls can be seen. While this is the first report of the antibacterial properties of GANand PEG-crosslinked LIG hydrogels, LIG has previously demonstrated promising antimicrobial activities towards cultures of Gram-positive bacteria, including Listeria monocytogenes and S.
aureus, and yeast, including Candida lipolytica 26,27 . The antimicrobial properties of lignin have previously been attributed to the polyphenolic components of this compound, which are reported to damage microbial cell membrane integrity with resultant bacterial lysis 28,29 . With respect to the S11 activity of LIG towards Gram-negative bacteria, mixed findings have been reported.  h incubation at 37°C. Columns and error bars represent means ± standard deviations (n ≥ 5).